DESCRIPTION: This application from the SGHMS forms part of a cooperative trial of 8 institutions to study over 1400 patients with 4 different tumors (soft tissue sarcomas, non-Hodgkins lymphoma, carcinomas of the head and neck, and locally advanced breast carcinoma). It will test the hypothesis that well-localized, in vivo 31-P MR spectra contain prognostic information that can predict response to treatment by correlating immediate response to metabolic features in the initial spectrum and to changes that occur in the spectra early following initiation of treatment. It will also test the hypothesis that the information from 31-P MRS can predict clinical outcome by correlating disease free survival in responders, and where feasible, overall survival to such metabolic features. This particular application is especially concerned with the development and testing of automated, objective pattern recognition techniques for the assessment of the 31-P tumor spectra produced at the collaborating centres to test the hypothesis that the patterns of metabolites revealed in the 31-P spectra of tumours can be correlated with clinical outcome, as described above. Techniques are or will be established at this, and all other interacting institutions, to include dual tuned 1H-31-P surface coils, proton decoupling of 31-P, 3D-CSI, molar quantitation, and pattern recognition for spectral analysis.